First delivered as a guest speaker at Mark Esposito's System & Complexity Thinking course at Hult University, June 2021.
This talk is aimed at people learning about complex systems with a background in organisational design & management. It looks at the difference between Computational Complexity and Applied Social Complexity, touches on the limits of Computation, Business Processes, & Systems Thinking, and introduces two frameworks that can be used to help organisations understand their current context and navigate more effectively.
Most content in the notes.
17. image: Brian Castellani, 2018 art-sciencefactory.com/complexity-map_feb09.html
Dave Snowden
“Anthro” Complexity
18. How does it fit in?
Source: Dave Snowden & Peter Stanbridge
“The landscape of management: Creating the context for
understanding social complexity” (2004)
21. R
R
legend
daughter me
Seek
Attention
Whinge R
Annoyed
Positive
Self-
Esteem
Fed Up
Punish
Happy
Praise
Play
Fun
R
✘
✘
✘
(Snowden & Stanbridge, 2004)
Systems thinking is a useful tool for making sense
of complex systems... But it only gives you the
perspectives you put in, and it assumes causality
where there may be none. Use it with that in mind.
26. It’s obvious
what to do.
Experts or due
process would tell us
the right thing to do.
Hindsight would tell
us the right thing to do, we
can’t know in advance.
This is random &
unexpected, an urgent
response is needed!
more detail: cynefin.io/wiki/4-points
27. Chaos
Act → Sense → Respond
Complex
Probe → Sense → Respond
Complicated
Sense → Analyse → Respond
Clear
Sense → Categorise → Respond
35. Evolution
Value
Chain
visible
invisible
Genesis Custom Product (+ rental) Commodity (+ utility)
uncharted industrialised
Modal’s
Customer
E-commerce Site
E-commerce
Platform as a
Service
(eg: SFCC,
Shopify)
competition
standards
competition
& standards
Hosting
(eg: AWS,
Azure,
Google
Cloud)
E-commerce
Framework
User Interface
Customisations
future
competition
Third Party
Integrations
User Experience
& Interface
Design
product
add-ons
Bespoke
Customisations
E-commerce
Consulting
Technical
Support
Modal
Digital’s
Value
Chain
Already
under
threat!
Inertia
36.
37. Evolution
Value
Chain
visible
invisible
Genesis Custom Product (+ rental) Commodity (+ utility)
uncharted industrialised
Modal’s
Customer
E-commerce Site
E-commerce
Platform as a
Service
(eg: SFCC,
Shopify)
Hosting
(eg: AWS,
Azure,
Google
Cloud)
E-commerce
Framework
User Interface
Customisations
Third Party
Integrations
User Experience
& Interface
Design
Bespoke
Customisations
E-commerce
Consulting
Technical
Support
Modal
Digital’s
Value
Chain
COVID-19
forcing
function
✘
✘
39. Evolution
Value
Chain
visible
invisible
Genesis Custom Product (+ rental) Commodity (+ utility)
uncharted industrialised
Modal’s New
Customer
User Interface
Customisations
Third Party
Integrations
User Experience
& Interface
Design
Bespoke
Customisations
Technical
Support
Modal
Digital’s
Value
Chain
?
?
Exaptation
Decompose to
find novel uses
✘
✘
40. Evolution
Value
Chain
visible
invisible
Genesis Custom Product (+ rental) Commodity (+ utility)
uncharted industrialised
Coherent
Probes
Modal’s New
Customer
User Interface
Customisations
Third Party
Integrations
User Experience
& Interface
Design
Bespoke
Customisations
Technical
Support
Modal
Digital’s
Value
Chain
?
?
✘
✘
41. Organised Services Operating Model
vs
Functional Hierarchy
James
Duncan
Service
Manager
design, delivery,
operations,
finance,
...
Container Service
Service
Owner
Executive
strategy, governance
Security Scan
Service
Web Service
Sub-teams as
needed
Service
Contract
Compute Service
Service
Mandates
https://osom.guide/
44. Recommendations (see slides for other links)
Cynefin
● Cynefin introduction (8min video)
● A Leader's Framework for Decision Making (HBR
Article, 2007)
● Cynefin.io (practitioner’s wiki)
● Cognitive Edge
Apex Predator / Flexuous Curves
(not covered, but very topical)
● Evolving Naval Safety: An Apex Predator Approach
Wardley Maps:
● Humorous Wardley Maps intro (20min video)
● Book: medium.com/wardleymaps
● Steve’s Introduction to Wardley Maps, and example
of real-world use: Wardley Maps @ Glasswall
● LearnWardleyMapping.com
● Organised Services Operating Model Guide
Conferences
● International Conference on Complex Systems (ICCS)
● Map Camp 2021
● Complex Systems Society (caveat: I’ve not been yet)
Complexity & Systems Thinking (sites, pdf’s, videos)
● Complexity concept map (Yaneer Bar-Yam)
● Systems-thinking: A Little Film (12 min)
● John Boyd’s OODA Loop
● This simple equation will change how you see the world -
Veritasium (20min video)
Related books:
● Complex Adaptive Systems: An Introduction to
Computational Models of Social Life (Miller & Page)
● Chaos (James Gleik)
● Order out of Chaos (Ilya Prigogine)
● Knowledge Assets (Max Boisot)
● Cynefin® book (Dave Snowden et al)
Editor's Notes
I first started out with the topic of complexity when I was a kid...
If you remember back to your highschool days where they get you up in front of the class to do a speech? Well, I did one about Fractals, I just remember being amazed that you could create things that looked like coastlines no matter how much you zoomed in… and then, if you look at our coastline from space and zoom in, it’s the same thing! That got me interested in Chaos theory, finding order where there seemingly was none, and eventually to software agents and artificial neural networks.
Anyways, we’re here to talk complexity, not hear my life story ;-)
I want to talk about making sense of complex systems, and I don’t want this to be overly academic, so I’m going to throw in real-life examples to show how this stuff is really useful.
Now, it’s day 3 of your course, so Mark’s hopefully already explained what a complex system is ;-). To recap:
Well, it’s complicated.
Despite it being a relatively young field, there’s already some 50 different areas of complexity science highlighted in Brian Castellani’s diagram here.
My early introduction came from a computational complexity perspective, and that’s probably the one you’ve heard most about.
Made popular by The Santa Fe institute, and the New England Complex Systems Institute or NECSI, and more recently by the growing field of Big Data & Artificial Intelligence… you can think of computational complexity as agent based models and mathematical simulations.
At last year’s International Conference on Complex Systems, I was impressed to see how a number of people were using agent based models.
Aistis from EndCoronavirus.org used them to simulate the spread of covid-19. While we all hated being cooped up in lockdowns and wearing masks when we go out, models like this have literally saved millions of lives.
Sandra Chapman showed how her team was connecting magnetometers around the world to allow us to better predict when space weather will cause havoc with our power grids and IT systems. Forewarned is forearmed!
And Paolo Gaudiano’s organisation, Aleria, uses them to show the impact of gender bias. Aleria’s work shows how this problem is systemic, and even better, what to do about it. This diversity is critical to cultivate for complex social systems, as we’ll see in a moment.
I’ve used computational complexity myself... last year as part of our Hult Business Challenge, I helped a company called X-Zell to create a prototype of a deep learning algorithm to detect cancer in blood samples by highlighting certain rare cells to pathologists. They have gone on to productise this in their quest for zero deaths from cancer. There is clear social & business value in this!
I’ve also created computer games with basic NPCs that result in gameplay you couldn’t predict. There’s a big market there too.
While powerful, computational complexity has its limits:
First, it is deeply technical and so the barrier to entry is high. Data visualization techniques have really improved the accessibility of these models… we all know that a picture says a thousand words!
Second, and perhaps more importantly, it is limited by the extent to which the model actually represents reality. That in turn is limited by things like:
the rules & algorithms used by the models
the time & effort put into creating the models
the capabilities & biases of the people creating the models
the intentions & biases of the organisations creating the models
the data used to build & operate the models
and our ability to interpret the models
Computation offers a very valuable perspective, but it’s only one perspective. If I’ve learned anything, it’s that you need to actively seek out multiple diverse perspectives to make sense of & navigate complex adaptive systems.
As you probably know by now, it’s not just shoals of fish, birds, ants, or honeybees... Humans also form complex adaptive systems wherever we go. Our markets, economies, businesses, governments, charities, teams, friends and families all form complex adaptive social systems. These all show the telltale signs of complexity:
A high number of interacting agents
With emerging patterns of behaviour
That can’t be predicted
And adapt to changing circumstances
With non-linear responses
The complexity is in the connections between people, how we interact, our shared values & beliefs, our goals, and how we live day-to-day. In business terms, it’s reflected by our cultures, our vision & mission statements, our strategies & tactics, our policies & procedures, our reward systems, and our daily activities.
Consider this:
If you were on a team of 10 people, there’d be 100 interpersonal relationships.
Add in 6 stakeholders, and you’re at 256.
Trying to predict what any individual will do in a given circumstance is a complex problem, as we are each influenced by a myriad of factors not least of which is our interactions with one another and the environments we occupy
And that’s where applied social complexity comes in.
You might be thinking: “how does this fit in with the other stuff I’m learning this weekend?”
Well, I think this diagram does a good job of putting it into perspective. It plots management approaches on two axes:
Y: is the approach ordered or un-ordered?
X: is the approach based on concrete rules? Or is it based on practical judgement, heuristics (rules-of-thumb)
Note the placement of systems thinking, which assumes order. That’s because it assumes causality, and that’s a fundamental point of difference:
cause and effect cannot be known in advance in a complex system.
That’s right, complex systems are not causal.
At best, you can glean some relationship after the fact. But even that “retrospective coherence” can be tenuous and dangerous if taken as truth.
And that really grates on people, because we seek order and we fear uncertainty.
We want to know, and we want things to be simple. In fact, we’re wired for it, as Kahneman describes in Thinking Fast & Slow.
Don’t get me wrong, systems thinking is a really useful tool to make sense of complex systems. But my point is, like computational complexity, it only gives you the perspectives you put in, and it assumes causality where there may be none. You just need to interpret it in that context.
I highlight this, because this is one of the biggest stumbling blocks I see. I meet many business leaders who have grown up under Taylor’s principles of Scientific Management and expect their business to exist solely in highly-ordered systems that they can control. Here, KPIs for outcomes can be set because they are known in advance.
Sadly, this is only true some of the time.
In a complex adaptive system, Dave Snowden teaches us to set directional KPIs, as outcomes cannot be planned, and your actions may change the landscape invalidating what you thought you wanted.
You can think of these as experiments or probes.
The problem is, how do you know what kind of system you’re currently in?
Attending a course like this will certainly help you!
There are also two really useful tools for this kind of situational awareness.
The first sense-making tool is Dave Snowden’s Cynefin framework.
Getting started is really simple.
Next time you’re feeling stuck, grab a colleague and put these 4 sentences up on the corners of a whiteboard. Then write whatever it is you’re thinking about on a sticky note, and discuss where it fits.
Think of the corners as attractors. It’s ok if your sticky gets stuck halfway between different corners, that’s normal. The point is to think about it and use your judgement to find the right place.
When you’ve done that, you can add in the Cynefin framework back in, and use it to help you decide what to do:
If you find yourself in Complicated or Clear, then reach for those familiar outcome-based KPIs and detailed project plans.
If you find yourself in Complex, think about what experiments or probes you should run in parallel.
If you’re in Chaos, well, put out the fire before doing anything else!
It gets more nuanced than that, of course, but even this level of usage can help you make better decisions!
The second sense-making tool is Simon Wardley’s value chain mapping technique, and it may be a bit reminiscent to Systems Thinking.
Created by Simon Wardley over a decade ago because he found the prevailing business strategy tools lacking.
Briefly, wardley maps are:
A chain of Needs
Drawn from the perspective of a User (more visible @ top)
Each link represents some flow of value (eg: needs met, finance, energy, risk, CO2, etc.)
With the added dimension of Evolution (newer @ left)
One of the most useful things about a map is that we can then have a conversation about it
You might disagree with me, for example:
Homemade Mugs are not “Custom”
Sure, the design your daughter made was custom
But the actual mug, paints, and baking process is a product
These discussions help groups of people examine multiple perspective to make sense of things.
Here’s a quick example, of a real company’s journey over the past year.
Modal Digital is an Ecommerce Solutions Integrator that, on the surface, looked well-equipped to weather the storm
(But underneath it all, were actually suffering without really knowing why)
When you map, you realise they are suffering from inertia.
Modal resisted the shift to Ecommerce platforms. Why?
Successful business
Using PaaS meant lower revenues: No hosting or support? Greatly reduced need for bespoke? No thanks!
Plus, what would we do with the capability we’ve built up?
These market shifts had early signals in 2015, and became obvious in 2016-2017.
Unfortunately, they experienced the Covid forcing function.
Accelerating the change.
Their customers still needed to sell
But they weren’t looking to spend a lot
It’s loss mitigation behaviour, and the platforms offer a better deal
Modal is one of many organisations that were forced over the cliff of complacency
The unfortunate ones go extinct,
The lucky ones like Modal took quick action to stop the bleeding.
And found themselves in a complex situation they didn’t fully understand.
The sensible ones cast about with multiple probes that may:
Prompt exaptation, reusing existing capabilities for things they weren’t really meant for (Liz’s dinosaurs + feathers example)
Or they may spark a process of adaptation.
------------------
exaptation: reusing capabilities for purposes they weren’t really planned for
adaptation: produced by natural selection for its current function
Exaptation can be the quickest route to recovery, and these Maps can help us do that.
They can help us breakdown our capabilities and look for novel uses, so we can recombine them into new products & services.
You can use maps like this to do scenario planning. They act as coherent probes which you can use to find new opportunities.
BTW: Modal found their feet by downsizing and switching focus to User Experience.
These maps can also be used to create adaptive organisational structures.
The best example I’ve seen of this is Stance’s “Organised Services Operating Model”.
Rather than using a classic functional hierarchy
OSOM breaks up organisations into cells that deliver a Service
Each cell has a contract which details the services provided, and a service level agreement
This doubles as a fitness function to evaluate performance of the service
The Executive decides which services to spin up, which to spin down.
It also governs via mandates, rather than dictats
This is all based on Wardley Maps: in fact, it’s derived from Simon’s Universal Doctrine.
At any rate, I’ve got loads more examples I could show you, but I don’t want to go too deep without leaving any time for discussion & questions.
So to sum up:
There’s 2 main approaches I’ve seen to complex systems: computation, and applied social
Seeking out multiple diverse perspectives are key to making sense of and navigating complex adaptive systems
Focus on increasing your situational awareness
Tools like Systems Thinking, Wardley Maps, and Cynefin can help you do that
If you found this session useful, or if you have any feedback please let me know!